Create single-evaluation

pages/single-evaluation

@@ -0,0 +1,80 @@
+import pandas as pd
+import streamlit as st
+from transformers import pipeline
+import os
+
+# Load models
+model1 = pipeline("text-classification", model="vectara/hallucination_evaluation_model")
+model2 = pipeline("text-classification", model="sileod/deberta-v3-base-tasksource-nli")
+
+# Predefined examples
+examples = {
+    'good': {
+        'question': "What causes rainbows to appear in the sky?",
+        'explanation': "Rainbows appear when sunlight is refracted, dispersed, and reflected inside water droplets in the atmosphere, resulting in a spectrum of light appearing in the sky.",
+        'ground_truth': "Correct"
+    },
+    'bad': {
+        'question': "What causes rainbows to appear in the sky?",
+        'explanation': "Rainbows happen because light in the sky gets mixed up and sometimes shows colors when it's raining or when there is water around.",
+        'ground_truth': "Incorrect"
+    }
+}
+
+# Function to evaluate explanations using the two models
+def evaluate_explanation(question, explanation):
+    results1 = model1(explanation)
+    results2 = model2(explanation)
+    return results1, results2
+
+# Function to compare vectors (simple difference in scores as example)
+def compare_vectors(v1, v2):
+    diff = abs(v1[0]['score'] - v2[0]['score'])
+    return diff
+
+# Title of the application
+st.title('Dual Model Evaluation of Explanations')
+
+# Check for password before allowing access
+def check_password():
+    def password_entered():
+        if password_input == os.getenv('PASSWORD'):
+            st.session_state['password_correct'] = True
+        else:
+            st.error("Incorrect Password, please try again.")
+
+    password_input = st.text_input("Enter Password:", type="password")
+    submit_button = st.button("Submit", on_click=password_entered)
+
+    if submit_button and not st.session_state.get('password_correct', False):
+        st.error("Please enter a valid password to access the demo.")
+
+# Password check
+if not st.session_state.get('password_correct', False):
+    check_password()
+else:
+    st.sidebar.success("Password Verified. Proceed with the demo.")
+
+    input_type = st.radio("Choose input type:", ('Use predefined example', 'Enter your own'))
+    if input_type == 'Use predefined example':
+        example_type = st.radio("Select an example type:", ('good', 'bad'))
+        selected_example = examples[example_type]
+        question = selected_example['question']
+        explanation = selected_example['explanation']
+        ground_truth = selected_example['ground_truth']
+    else:
+        question = st.text_input('Enter your question:', '')
+        explanation = st.text_input('Enter your explanation:', '')
+        ground_truth = st.text_input('Enter ground truth:', '')
+
+    if st.button('Evaluate Explanation'):
+        if question and explanation and ground_truth:
+            results1, results2 = evaluate_explanation(question, explanation)
+            diff = compare_vectors(results1, results2)
+            st.write('### Model 1 Results')
+            st.write(results1)
+            st.write('### Model 2 Results')
+            st.write(results2)
+            st.write(f'### Score Difference: {diff}')
+        else:
+            st.error('Please enter a question, explanation, and ground truth to evaluate.')